AutoCA
Custom AI invoice extraction, GST auto-reconciliation, WhatsApp document collection, and one-click Tally export — built for India's CA ecosystem.
Available for new projects — 2026
Enterprise AI
Engineered to Scale.
I build production-grade LLM systems, AI agents, and document intelligence pipelines — and I'm the founder of Sahaj, creator of AutoCA, India's AI invoice platform for CAs and tax professionals.
50+
Projects shipped
5+ yrs
AI engineering
98%
Client satisfaction
Google
+ Azure certified
What I build
AI systems that replace
manual workflows at scale.
LLM Application Development
End-to-end LLM apps using GPT-4, Claude, Gemini, or open-source models — with RAG, tool use, and memory systems built for production.
RAG Pipeline
Fine-tuning
Tool Use
Prompt Eng.
Agentic AI Systems
Multi-step AI agents that reason, plan, and act across APIs, databases, and external tools — from research agents to automated business workflows.
LangChain
LangGraph
AutoGen
MCP
OCR & Document Intelligence
Convert unstructured documents — invoices, bank statements, contracts — into structured data with AI-powered extraction, validation, and automation.
Invoice OCR
PDF Parsing
Validation
GST Recon
Enterprise AI Integrations
Connect AI to your existing stack — Salesforce, SAP, ERPs, CRMs — using MCP servers, custom APIs, and event-driven pipelines. Zero disruption.
Salesforce
Agentforce
REST/gRPC
Webhooks
AI SaaS Product Development
Full-stack AI product builds — from architecture to deployment. React frontend + FastAPI backend + LLM layer + cloud infra, delivered as a working SaaS.
React
FastAPI
Docker
GCP/Azure
AI Audit & Strategy Consulting
AI readiness assessment, stack evaluation, and a 90-day roadmap for enterprise AI adoption. Know exactly where AI will have the highest ROI in your business.
AI Roadmap
ROI Analysis
Architecture
Team Training
Case studies
Production AI systems.
Measurable business outcomes.
01
5k+
Practicing CAs
12k+
Tax professionals
Problem
CA and tax firms lose 20+ hours weekly to manual invoice entry, GST mismatch checks, and chasing clients for documents — especially during filing season when volume spikes and errors are costly.
Solution
AutoCA is India's custom AI invoice automation platform: upload any invoice (image, PDF, Excel), proprietary extraction models parse every field, auto-reconcile GST, and export to Tally, Zoho, or QuickBooks in one click — plus WhatsApp document collection from clients.
Product highlights
Custom AI Models
GST Reconciliation
Tally Export
WhatsApp Collect
Data Entry Automation
GSTR Filing Ready
02
95%
Extraction accuracy
10x
Faster processing
Problem
NBFCs and lending platforms needed to analyse bank statements from 40+ banks in different formats to determine creditworthiness — a process that took 2–3 days per applicant and was bottlenecking loan disbursals.
Solution
Designed a multi-bank OCR pipeline that parses statements from any format, categorises transactions, detects suspense entries, calculates EMI obligations, and generates a structured credit analysis report — end to end in under 60 seconds.
Tech Stack
PaddleOCR
LLM Extraction
Pydantic
FastAPI
Redis
Docker
Azure
03
360°
Customer view
0 code
For end users
Problem
Sales reps needed to query Salesforce CRM data, update records, and generate reports — tasks that required navigating complex UIs or waiting for BI teams. No natural language interface existed for real-time CRM intelligence.
Solution
Built a Claude AI agent connected to Salesforce via an MCP server layer, enabling natural language queries against live CRM data, Agentforce-powered actions, and headless 360-degree customer views — all through a chat interface.
Tech Stack
Claude API
MCP Protocol
Salesforce API
Agentforce
Headless 360
FastAPI
Node.js
Portfolio
Selected AI products
shipped to production.
Client work plus AutoCA — my live SaaS for CA firms. End-to-end from architecture to production.
My product · Live at autoca.in
AutoCA — AI Invoice Automation for CAs & Tax Professionals
Upload any invoice — image, PDF, or Excel. Custom extraction models parse every field, reconcile GST, and push to Tally in one click. Trusted by 5,000+ practicing CAs and 12,000+ tax professionals.
20+ hrssaved per practice / week
1-clickTally · Zoho · QuickBooks
7-dayfree trial
AI Finance App
Real-time and historical market analysis with LSTM models, Yahoo Finance / Polygon.io feeds, and FastAPI backends on GCP.
AI Astrology
Personalized readings and scheduling insights using NLP/LLM pipelines and transformer models with explainable outputs.
AI Travel Agent
Retrieval-augmented itineraries with multi-source search — on-budget, preference-aware trip plans via conversational workflows.
Object Detection
Real-time TF2 object detection pipeline for vision workloads — optimized inference and integration with cloud-ready APIs.
AI Trading Bot
LangChain agents with vector search and market feeds for conversational stock/crypto analysis and portfolio alerts.
Image Generation
Mobile GenAI experience for AI-powered image creation — optimized UX for generation, preview, and export workflows.
UI/UX Design
Interfaces built for
clarity, speed & trust.
Product design for AI dashboards, chat experiences, mobile apps, and conversion-focused landing pages — aligned with your brand and user workflows.
Dashboard · Finance
AutoCA Dashboard
Live invoice hub — totals, pending queue, drag-and-drop upload, and voucher-ready Tally export. autoca.in
Dashboard · Lending
Credit Analysis Console
Underwriter view with transaction categories, EMI detection, and exportable credit memos.
Chat UI · CRM
Salesforce AI Assistant
Conversational CRM interface with live data citations and inline record actions.
Mobile · FinTech
Loan Status Tracker
Applicant-facing app for document upload, verification status, and disbursement timeline.
Landing Page · SaaS
AI Product Marketing Site
Conversion-focused hero, social proof, feature grid, and Calendly-integrated CTA flow.
Chat UI · RAG
Knowledge Base Copilot
Citation-backed answers with source previews and feedback loops for continuous improvement.
Figma
Framer
Design Systems
Wireframing
Prototyping
User Flows
Accessibility (WCAG)
Responsive Design
How it works
From first call to production in weeks, not months.
01 / Discovery
Deep business audit
45-min discovery call. I map your workflows, identify AI opportunities, and quantify the ROI of each. You get a clear picture of what's worth building.
02 / Architecture
System design & scoping
I design the AI architecture — model selection, data flows, APIs, infra — and deliver a detailed scope with milestones, stack choices, and cost estimates.
03 / Build
Rapid iterative development
Weekly demos, async updates, and staging previews. Production-grade code with unit tests, error handling, logging, and cloud deployment from day one.
04 / Deploy & Scale
Handoff, training & support
Full deployment, team training, and documentation. 60-day post-launch support window. Optional retainer for ongoing feature development and monitoring.
Tech stack
Production tools.
Proven at scale.
Best-in-class AI, cloud, and design tooling — selected for reliability, observability, and long-term maintainability.
Expert
OpenAI GPT-4 / o1
RAG, agents, structured outputs
Expert
Claude (Anthropic)
MCP, long-context, tool use
Expert
LangChain / LangGraph
Multi-agent orchestration
Advanced
LlamaIndex
Document indexing & retrieval
Advanced
Google Gemini
Multimodal & enterprise APIs
Advanced
Ollama / Local LLMs
On-prem & cost-sensitive workloads
Expert
Python
AI pipelines, APIs, automation
Expert
FastAPI
High-performance async APIs
Advanced
Google Cloud
Vertex AI, Cloud Run, GCS
Advanced
Microsoft Azure
OpenAI Service, AKS, Blob
Advanced
Amazon Web Services
Lambda, S3, Bedrock
Expert
Docker & Kubernetes
Containerised AI deployments
Expert
Pinecone
Managed vector search at scale
Advanced
Weaviate
Hybrid search & filtering
Expert
FAISS
Local / embedded vector stores
Expert
PostgreSQL
pgvector, transactional data
Advanced
Redis
Caching, queues, sessions
Advanced
MongoDB
Flexible document stores
Expert
React
AI dashboards & SaaS UIs
Expert
Next.js
SSR, SEO-friendly AI products
Expert
Tailwind CSS
Rapid, consistent UI systems
Expert
TypeScript
Type-safe frontends & SDKs
Advanced
Vite
Fast dev & production builds
Advanced
WebSockets / SSE
Streaming LLM responses
Expert
Figma
UI design, prototypes, handoff
Advanced
Framer
Interactive marketing sites
Expert
Design Systems
Tokens, components, patterns
Advanced
Accessibility (WCAG)
Inclusive AI product UX
Expert
User Flows & Wireframes
Discovery to high-fidelity
Expert
Chat & Dashboard UX
AI-native interface patterns
Client feedback
What clients say.
"Finished the AI trading bot project with great expertise, understood our problem immediately and delivered on time. Helped us with clear requirements from the start. We want to continue with further tasks."
"Sahaj is very talented. He delivered before time, was always available to give updates, and has a great cooperative nature. I will definitely hire him again for upcoming projects."
"Super professional and always responsive. He did exactly what we needed with no fuss. I would highly recommend Sahaj for any AI project — excellent to work with."
SG
Sahaj Godhani
Founder of Sahaj · Surat, India
Available
About me
GenAI Consultant & Product Builder.
Founder of Sahaj
I've shipped AI systems for finance firms, lending platforms, enterprise CRM users, and SaaS founders — across India, Europe, and the US. I care deeply about one thing: does it actually work in production, and does it move the business metric it was built for?
G
Prompt Design in Vertex AI
Google Cloud · 2024
Az
Azure AI Fundamentals
Microsoft · 2024
What industries do you work with?
My core focus areas are finance & CA firms (invoicing, GST, compliance automation), banking & lending (document processing, credit analysis), SaaS products (AI feature development), and enterprise companies needing Salesforce or CRM AI integrations. I also work with ecommerce brands and general SMBs automating repetitive knowledge work.
How long does a typical project take?
A focused AI automation module (e.g. invoice OCR pipeline) takes 3–4 weeks. A full AI SaaS product build runs 8–14 weeks. Enterprise integrations and agentic systems fall between 4–8 weeks depending on API complexity. I'll give you a precise timeline during scoping.
Do you work with open-source models or only GPT/Claude?
Both. For most enterprise clients, GPT-4 or Claude is the right choice due to reliability and capability. For clients with data privacy requirements or cost constraints, I build with Llama 3, Mistral, or other open-source models hosted on your own infrastructure. Model selection is always ROI-driven, not brand-driven.
What does your engagement look like post-launch?
Every project includes a 60-day support window for bug fixes and minor adjustments. After that, I offer optional retainer arrangements for ongoing development, monitoring, and feature iteration. Many clients work with me on a retainer basis after the initial build.
What's your pricing structure?
I work on project-based fixed pricing for well-scoped builds, and time-and-material for exploratory or evolving projects. Discovery calls are always free. Pricing is shared after the scoping session, when I have a clear picture of what needs to be built. Projects typically range from ₹1.5L for focused automation modules to ₹8L+ for full-stack AI products.
Is my data safe when you build AI systems?
Data security is built into the architecture from day one. I can design systems that keep sensitive data fully within your cloud environment. For regulated industries (banking, CA), I implement data masking, encryption at rest and in transit, audit logging, and PII handling that meets compliance standards. NDAs are standard practice.
Get in touch
Let's build something
that actually works.
Book a free 30-minute discovery call. I'll map your workflow, identify the highest-ROI AI opportunities, and give you an honest assessment — no sales pitch.
Email
sahajgodhani777@gmail.com
Phone / WhatsApp
+91 74051 40523
Location
Surat, Gujarat, India
Send a message
I respond to every serious enquiry within 24 hours.
Or book directly: Schedule a 15-min call →
Ready to start?
Your AI system is
8 weeks away.
Stop watching competitors automate while your team does it manually. Book a free discovery call — I'll show you exactly what's possible, and what it'll cost.